Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Embedding Propagation: Smoother Manifold for Few-Shot Classification
Pau Rodríguez, Issam Laradji, Alexandre Drouin, Alexandre Lacoste
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add embedding propagation to label propagation (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 53.40 -> 55.75 | Improved |
| add embedding propagation alone (prototypical classifier) (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 49.57 -> 52.83 | Improved |
| add label propagation to embedding propagation (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 52.83 -> 55.75 | Improved |
| add episodic fine-tuning (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 49.57 -> 54.29 | Improved |
| add rotation loss (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 49.57 -> 50.83 | Improved |
| add embedding propagation on top of fine-tuning, rotation loss and LP (ablation) | EPNet · miniImageNet 1-shot (Conv-4) | accuracy (%): 57.35 -> 58.85 | Improved |
| add embedding propagation (EPNet vs -Net) (ablation) | EPNet with WRN-28-10 · miniImageNet 1-shot | accuracy (%): 65.98 -> 70.74 | Improved |
| add embedding propagation (EPNet vs -Net) (ablation) | EPNet with ResNet-12 · miniImageNet 5-shot | accuracy (%): 81.28 -> 81.06 | No clear effect |
| add embedding propagation (EPNet vs -Net) (ablation) | EPNet with Conv-4 · miniImageNet 5-shot | accuracy (%): 72.57 -> 72.95 | No clear effect |
| add embedding propagation to wDAE-GNN | wDAE-GNN · Imagenet-FS novel classes K=1 (top-5) | accuracy (%): 48.00 -> 50.07 | Improved |
| add 100 unlabeled samples (semi-supervised) | EPNet with ResNet-12 · miniImageNet 1-shot | accuracy (%): 66.50 -> 75.36 | Improved |
| remove off-diagonal of propagator (keep diagonal only) (ablation) | EPNet with ResNet-12 pre-training · miniImageNet 5-shot | accuracy (%): 73.80 -> 75.95 | Improved |
Closest papers by meaning
- A Closer Look at Embedding Propagation for Manifold Smoothing (JMLR 2022 · similarity 0.94)
- LEARNING TO PROPAGATE LABELS: TRANSDUCTIVE PROPAGATION NETWORK FOR FEW-SHOT LEARNING (ICLR 2019 · similarity 0.88)
- A Study of the Classification of Low-Dimensional Data with Supervised Manifold Learning (JMLR 2018 · similarity 0.87)
- Zero Shot Learning with the Isoperimetric Loss (AAAI 2020 · similarity 0.87)
- Euclidean Embedding of Data Using Local Distances (NeurIPS 2026 · similarity 0.85)
- Boosting Few-Shot Learning via Attentive Feature Regularization (AAAI 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
